In slopp/renv: Project Environments for R

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

If you're using renv with an R project that also depends on some Python packages (say, through the reticulate package), then you may find renv's Python integration useful.

Activating Python Integration

Python integration can be activated on a project-by-project basis. Use renv::settings$python(TRUE) to tell renv to create and use a project-local Python virtual environment with your project. If the reticulate package is installed and active, then renv will use the same version of Python that reticulate normally would when generating the virtual environment. Alternatively, you can set the RETICULATE_PYTHON environment variable to instruct renv to use a different version of Python.

If you'd rather tell renv to use an existing Python virtual environment, you can do so by passing the path of that virtual environment instead -- use renv::settings$python("/path/to/python/virtualenv") and renv will record and use that Python virtual environment in the project.

You can also pass the name of a virtual environment (e.g. renv::settings$python("tensorflow")) to use the virtual environment called "tensorflow" located at ~/.virtualenvs/tensorflow.

Understanding Python Integration

Once Python integration is active, renv will attempt to manage the state of your Python virtual environment when snapshot() / restore() is called. In particular, renv does the following:

• renv::snapshot() calls pip freeze > requirements.txt to save the set of installed Python packages;

• renv::restore() calls pip install -r requirements.txt to install the previously-recorded set of Python packages.

With this, projects that use renv and Python can ensure that Python dependencies are tracked in addition to R package dependencies. Note that future restores will require both renv.lock (for R package dependencies) and requirements.txt (for Python package dependencies).

slopp/renv documentation built on July 6, 2019, 12:08 a.m.